R. Pitt March 8, 2011 The National Stormwater Quality Database, Version 3.1 The characteristics of stormwater discharges vary considerably. Geographical area and land use have been identified as important factors affecting base flow and stormwater runoff quality, for example. Many studies have investigated stormwater quality, with the EPA’s Nationwide Urban Runoff Program (NURP) being the best known and earliest effort to collect and summarize these data. Unfortunately, NURP was limited in that it did not represent all areas of the US or all important land uses. More recently, the National Stormwater Quality Database (NSQD) compiled runoff characteristics information from more than 8,000 events from throughout the US. Most of these data were from the EPA’s NPDES stormwater permit program for Phase 1 communities. These permits are needed for all large municipal areas having >250,000 in population. The Phase II permit program requires permits from small communities. As a condition for these permits, municipalities were required to establish a monitoring program to characterize their local stormwater quality for their most important land uses discharging to the municipal separate storm sewer system (MS4). Although only a few samples from a few locations are required each year from these communities, the ten plus years of MS4 data included in the NSQD comprise a suitable number of samples from many locations. Recently, version 3 of the NSQD was completed, and besides expanding to include additional stormwater NPDES MS4 permit holders, most of the older NURP data, and some of the International BMP database information was also added, along with data from some USGS research projects. A number of land uses are represented in these data, with most data from residential, commercial, and industrial areas, and less data from freeways, institutional and open space areas. These observations were all obtained at outfall locations and do not include snowmelt or construction erosion sources. This version contains the results from about one fourth of the total number of communities that participated in the Phase I NPDES stormwater permit monitoring activities. The database is located at: http://unix.eng.ua.edu/~rpitt/Research/ms4/mainms4.shtml. Figure 1 is a map showing the EPA Rain Zones in the US (not to be confused with EPA administrative regions), along with the locations of the communities containing data in the NSQD, ver 3. Recent revisions to the database have included additional quality control evaluations. In the near future, additional supplemental data from under-represented regions (especially southern California) will also be added to the database. Tables 1 through 7 summarize the data observations for selected stormwater characteristics (volumetric runoff coefficient, TSS, TKN, TP, Cu, Zn, and fecal coliforms). These data are separated by the six land used represented and geographical area (shown by EPA Rain Zones). Rain Zones 8 and 9 have very few samples, and institutional and open space areas are poorly represented. However, residential, commercial, industrial, and freeway data are plentiful, except for the few EPA Rain Zones noted above. The yellow high-lighted cells indicate rain zone-land use combinations having at least 40 events represented, a value expected to result in more reliable concentration estimates than for conditions having very few data. 1 Figure 1. Sampling Locations for Data Contained in the National Stormwater Quality Database, version 3.1. The data shown on Tables 1 through 7 are for comparison to the modeled conditions representing the standard land use files. The values on these tables are the averages, the coefficient of variation, and the number of observations. Besides each land use and rain zone combination, overall land use and overall rain zone values are also shown along with the overall database values. The average values are shown instead of the median, as the averages better represent long-term mass discharges. Median values artificially reduce the effects of the periodic unusually high concentrations that do occur in stormwater. Figure 2 is a plot showing the calculated ratios of the average concentrations to the median concentrations for all of the cells represented in the data set summarized in the tables. The average values are all larger than the median values, so the ratios range from about 1 up to about 10. This plot shows how this ratio generally increases as the coefficient of variation (COV) values increase. The coefficient of variation is the ratio of the standard deviation to the average value (another reason why the average values are shown on these tables). 2 Average to Median Concentration Ratio Stormwater concentrations usually have a log-normal distribution, resulting in a positive bias, resulting in the average values being larger than the median values. The greater the difference, the greater the positive bias (and the larger the COV). If the COV is less than about 0.5, there is little difference between the median and the average values. However, most of the stormwater concentration COV values are in the range of 0.5 to 1.5, as indicated on Figure 2, with some much larger. The bacteria observations have the largest variations in each sample subgroup, while the Rv and TKN have the smallest variations. 100 y = 0.90e0.53x R² = 0.71 10 1 0.1 0.1 1 10 Coefficient of Variation (COV) Figure 2. Relationship between the average to median concentration ratios and the COV values. In most cases, the COV values are slightly smaller for the subgroups compared to the overall group values, indicating that the land use and geographical combinations help explain some of the large variability commonly found with stormwater concentrations. Detailed analyses have been conducted using the complete database to statistically identify significant subgroupings. In most cases, the complete matrix of 54 combinations is reduced by missing data and by combining similar conditions, resulting in many fewer significant subgroups. The following tables do not reflect these statistical groupings, as these data are being used to verify the modeling calculations for specific locations. Actual data for the areas closest in character to the standard land use file locations are desired for these comparisons; therefore, the data in high-lighted cells are compared to the calculated values. If a cell is not high-lighted, then the land use high-lighted value is used. If that is not high-lighted, then the overall value is compared. 3 Table 1. Volumetric Runoff Coefficients, Rv, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial n/a 0.59 (0.5) 66 0.32 (0.7) 14 n/a n/a 0.31 (0.3) 37 n/a 0.59 (0.5) 16 n/a 0.28 (0.6) 9 0.43 (1.0) 54 0.36 (0.2) 7 n/a n/a Institutional n/a 0.04 (1.8) 14 0.34 (0.7) 50 n/a 0.67 (0.6) 158 0.34 (0.9) 69 0.48 (0.1) 2 n/a Industrial 0.68 (0.5) 114 0.46 (0.3) 20 0.72 (0.2) 110 n/a 0.65 (0.4) 34 Freeways 0.59 (0.9) 64 n/a n/a n/a n/a 0.30 (0.8) 23 n/a Open Space n/a 0.15 (0.6) 16 n/a 0.16 (1.1) 9 0.06 (0.5) 2 n/a n/a Residential 0.30 (2.4) 88 0.28 (1.7) 403 0.34 (0.6) 30 0.20 (0.7) 51 all land uses 0.30 (2.3) 97 0.32 (1.4) 705 % detect 100% 100% 0.18 (1.2) 209 0.28 (1.2) 322 100% 0.33 (0.8) 69 0.33 (0.7) 184 0.50 (0.6) 497 100% 0.23 (1.2) 30 0.28 (1.4) 37 100% 0.55 (1.0) 5 0.50 (1.3) 8 100% 0.62 (0.6) 310 0.58 (0.7) 215 0.48 (0.7) 322 0.04 (1.8) 14 0.29 (0.9) 103 0.27 (1.5) 1054 0.39 (1.1) 2115 n/a n/a 0.31 (0.6) 60 100% 0.51 (0.8) 314 100% 0.30 (0.6) 7 0.24 (1.0) 54 0.31 (0.9) 100 100% % detect 100% 100% 100% 100% 100% 100% 100% Table 2. TSS Concentrations, mg/L, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 201 (1.5) 310 24 (0.3) 3 101 (1.7) 669 56 (2.0) 55 232 (1.9) 67 132 (1.0) 41 87 (0.9) 61 98 (0.8) 7 80 (1.6) 225 36 (1.4) 13 n/a 108 (1.6) 100 144 (1.1) 12 n/a n/a 177 (1.4) 100 91 (0.7) 8 97 (1.6) 375 164 (1.4) 68 n/a n/a 155 (1.7) 106 n/a 164 (1.2) 30 86 (1.0) 46 105 (1.2) 105 68 (1.4) 15 183 (2.8) 105 385 (1.2) 95 247 (1.2) 32 n/a n/a n/a n/a 360 (0.9) 39 n/a 133 (1.7) 1342 114 (2.5) 381 160 (1.6) 918 83 (1.0) 69 176 (2.4) 128 135 (1.2) 507 156 (1.6) 1132 99% 98 (1.5) 107 n/a 370 (0.8) 18 202 (1.6) 67 330 (n/a) 1 n/a n/a 846 (0.4) 7 102 (1.7) 1893 97 (1.7) 3468 102 (1.6) 207 93 (1.6) 395 98% 129 (0.9) 203 141 (1.5) 488 99% 162 (1.0) 75 99% 374 (1.8) 140 293 (1.8) 293 99% 130 (1.8) 315 126 (1.7) 443 100% 140 (0.9) 16 140 (1.0) 24 100% 528 (2.5) 116 460 (2.3) 194 99% Freeways Industrial Institutional Open Space Residential all land uses % detect 235 (1.7) 318 90% 182 (1.9) 329 137 (2.4) 3472 135 (2.2) 6682 % detect 98% 100% 97% 99% 98% 99% 99% 4 Table 3. Total Phosphorus Concentrations, mg/L, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 0.25 (2.2) 311 0.43 (0.5) 3 0.37 (1.3) 641 0.38 (1.6) 50 0.34 (0.7) 16 n/a n/a n/a 0.35 (1.3) 84 0.35 (0.6) 24 0.33 (0.9) 76 n/a 0.57 (0.6) 7 n/a 0.36 (1.6) 370 0.64 (3.0) 112 0.22 (0.7) 245 0.25 (1.2) 108 n/a 0.57 (0.7) 37 0.33 (0.8) 100 0.21 (0.4) 8 0.46 (0.7) 23 n/a 0.18 (1.7) 139 0.40 (1.1) 565 0.32 (0.4) 1203 97% 0.33 (1.1) 106 0.39 (1.1) 141 0.16 (0.7) 14 0.20 (0.9) 108 0.19 (0.5) 15 n/a 0.65 (0.3) 2 n/a n/a 0.43 (1.7) 1956 0.42 (1.7) 3572 97% 0.20 (1.4) 410 0.24 (1.3) 688 95% 0.70 (1.2) 91 0.40 (1.0) 67 0.47 (0.9) 206 0.38 (2.2) 738 99% 0.54 (1.1) 70 0.30 (1.2) 331 0.31 (1.1) 539 99% 0.85 (0.7) 15 0.74 (0.8) 23 100% 0.37 (2.0) 1399 0.50 (1.7) 604 0.39 (1.5) 897 0.23 (0.17) 68 0.29 (1.2) 338 0.71 (1.5) 3719 0.40 (1.7) 7295 Freeways Industrial Institutional Open Space Residential all land uses % detect 0.95 (1.3) 186 0.24 (0.8) 45 n/a 0.36 (1.2) 49 n/a 0.31 (0.6) 17 0.51 (1.3) 207 98% 0.49 (1.6) 135 1.3 (0.9) 63 0.68 (1.3) 307 97% n/a 0.60 (0.5) 7 0.81 (1.1) 75 0.67 (1.1) 121 100% % detect 96% 99% 95% 99% 96% 98% 97% Table 4. Total Kjeldahl Nitrogen Concentrations, mg/L, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 1.5 (1.1) 185 2.0 (0.9) 625 1.2 (0.7) 41 1.8 (0.9) 47 4.3 (0.7) 39 1.6 (1.0) 61 3.7 (0.7) 5 Freeways 3.6 (0.3) 3 2.4 (1.1) 100 n/a n/a 3.3 (1.4) 122 1.7 (0.6) 24 n/a 2.6 (0.6) 16 n/a Industrial 1.9 (0.9) 100 1.8 (1.5) 338 1.5 (0.8) 99 1.6 (0.6) 46 4.2 (0.8) 76 1.9 (0.6) 33 n/a Institutional 0.79 (0.6) 7 1.6 (0.8) 46 1.4 (0.5) 15 n/a 1.1 (0.6) 112 2.0 (0.9) 204 1.2 (0.9) 109 n/a n/a n/a n/a 2.5 (0.6) 23 n/a 1.9 (0.9) 1131 2.4 (1.2) 450 1.9 (1.2) 824 1.5 (0.8) 68 Open Space 1.2 (0.8) 77 n/a 1.9 (0.7) 18 1.7 (0.9) 67 1.8 (0.2) 2 n/a n/a 3.3 (0.6) 7 Residential 0.79 (0.7) 100 1.9 (0.9) 434 1.8 (1.1) 1783 2.3 (1.5) 74 1.6 (0.9) 834 1.9 (1.1) 3067 % detect 100% 97% 2.1 (0.9) 183 1.7 (0.9) 675 96% 3.2 (2.7) 74 all land uses 1.0 (0.9) 335 1.2 (0.9) 490 93% 1.1 (0.9) 318 1.3 (0.9) 460 98% 5.7 (0.8) 15 5.0 (0.8) 21 100% 3.8 (0.7) 64 3.3 (0.7) 110 100% 2.0 (0.7) 185 97% 3.6 (1.0) 313 99% 1.3 (1.0) 271 1.8 (1.1) 3280 1.9 (1.1) 6095 % detect 97% 99% 96% 97% 91% 98% 97% 5 Table 5. Total Copper Concentrations, µg/L, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 58 (0.8) 141 33 (1.2) 502 69 (1.2) 47 61 (3.5) 109 21 (1.1) 40 29 (1.1) 84 42 (1.2) 7 46 (1.6) 32 Freeways 54 (0.1) 3 28 (2.0) 103 7.4 (1.4) 106 1.1 (0.5) 13 n/a 62 (1.4) 101 32 (0.8) 26 n/a n/a Industrial 25 (1.5) 83 22 (1.3) 257 18 (1.0) 106 99 (2.3) 49 7.4 (1.7) 117 17 (0.7) 107 78 (0.9) 93 42 (0.8) 34 n/a 46 (1.0) 39 Institutional 33 (0.4) 7 25 (0.7) 45 7.3 (0.6) 15 n/a n/a n/a n/a n/a n/a 37 (2.3) 1068 30 (2.0) 360 36 (2.0) 768 21 (0.8) 67 Open Space 9 (0.1) 6 9 (0.8) 58 n/a 20 (0.8) 12 12 (0.9) 70 119 (1.1) 2 n/a n/a 28 (0.7) 7 Residential 34 (1.8) 333 30 (1.6) 1340 10 (2.6) 396 52 (1.8) 111 16 (1.8) 164 36 (1.4) 66 13 (0.7) 24 22 (0.4) 15 all land uses 33 (1.6) 644 29 (1.5) 2339 10 (2.1) 636 65 (2.1) 219 23 (4.3) 567 56 (1.4) 302 26 (1.1) 253 28 (1.0) 23 % detect 78% 89% 79% 89% 98% 99% 93% 90% 28 (0.9) 103 35 (1.2) 181 83% 14 (1.5) 155 27 (1.8) 2613 30 (2.1) 5087 % detect 88% 98% 86% 85% 84% 88% 88% Table 6. Total Zinc Concentrations, µg/L, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 237 (1.2) 513 60 (1.4) 136 270 (0.9) 51 138 (1.0) 84 434 (1.1) 7 185 (1.3) 203 7.5 (0.9) 14 n/a 116 (0.8) 111 89 (1.2) 267 343 (2.0) 42 Freeways 196 (1.4) 225 368 (0.1) 3 304 (1.1) 99 211 (0.8) 25 n/a 217 (0.7) 32 n/a Industrial 106 (1.2) 84 172 (0.9) 326 512 (2.9) 54 254 (0.9) 46 n/a 1720 (2.0) 100 n/a n/a 169 (0.2) 7 169 (1.1) 107 n/a 306 (2.9) 81 Institutional 166 (1.3) 107 90 (0.5) 15 n/a n/a 486 (0.9) 39 n/a Open Space 53 (0.8) 10 93 (0.8) 109 n/a 98 (1.0) 17 100 (1.3) 69 225 (1.0) 2 n/a n/a 439 (0.4) 7 Residential 134 (1.2) 351 138 (1.4) 752 98% 125 (3.6) 1471 162 (2.3) 2711 97% 61 (1.2) 384 264 (2.3) 120 310 (2.7) 242 98% 95 (0.9) 183 260 (1.2) 76 107 (1.1) 737 100% 746 (2.8) 319 97% 120 (0.8) 328 152 (2.4) 542 100% 185 (0.6) 15 264 (1.1) 22 100% 139 (1.0) 100 242 (1.2) 178 94% 197 (1.4) 1201 159 (1.4) 608 382 (3.5) 898 210 (1.0) 68 109 (1.1) 214 125 (2.8) 3028 178 (3.3) 6036 all land uses % detect 78 (1.5) 656 95% % detect 99% 99% 99% 100% 91% 97% 97% 6 Table 7. Fecal Coliforms, count/100 mL, for Different Land Uses and Geographical Areas (EPA Rain Zones) (average, COV, number of observations) Land Use RZ1 RZ2 RZ3 RZ4 RZ5 RZ6 RZ7 RZ8 RZ9 all RZ Commercial 5,160 (2.2) 43 <1 (n/a) 3 3,220 (1.0) 6 n/a 54,500 (1.9) 44 n/a 21,600 (1.3) 11 n/a 83,400 (4.2) 46 n/a 34,000 (4.0) 57 7,060 (1.8) 23 15,100 (3.5) 34 n/a 3,500 (1.3) 4 n/a 5,500 (1.8) 65 n/a 41,000 (2.8) 103 9,000 (2.0) 13 50,200 (4.9) 109 n/a 9,500 (1.5) 7 100,000 (5.7) 44 3,100 (0.4) 3 14,200 (2.2) 154 11,400 (3.3) 18 14,000 (2.5) 150 n/a n/a n/a 24,200 (1.8) 15 n/a 14,300 (1.7) 6 210,000 (3.3) 156 140,000 (4.2) 301 80% 11,100 (3.1) 24 33,100 (2.7) 380 24,100 (2.9) 731 88% n/a 17,900 (1.0) 16 41,700 (1.0) 91 52,600 (3.4) 197 87% 39,900 (2.1) 67 88,500 (1.8) 165 59,300 (2.8) 457 100% 2,500 (n/a) 1 5,970 (1.7) 10 6,520 (1.9) 66 97% n/a n/a 480 (0.6) 2 25,400 (2.8) 68 25,000 (3.6) 191 95% 17,800 (2.1) 10 13,700 (2.3) 14 100% 25,600 (1.0) 8 22,400 (1.5) 36 100% 27,400 (3.2) 429 8,600 (2.5) 80 35,900 (6.6) 485 3,100 (0.4) 3 29,100 (2.4) 116 69,600 (4.4) 978 48,400 (5.0) 2102 Freeways Industrial Institutional Open Space Residential all land uses % detect 20,300 (6.5) 90 13,700 (7.2) 161 91% 7,900 (2.1) 26 4,190 (1.0) 22 n/a % detect 91% 100% 90% 100% 97% 91% 91% 7 During her Ph.D. research, Bochis (2010) examined all 2-way interactions between the geographical regions and the land use categories for selected constituents in the NSQD. She found that the national data could be combined into a reasonable number of significantly different subsets having similar characteristics. These groups of data have concentrations that are more similar within the group than between the groups. These groupings of the data can be used to assist local stormwater managers in estimating likely stormwater concentrations for similar local conditions. Examining 3-way interactions, by adding seasonal data to the geographical regions and land use information, did not result in many additional category distinctions associated with seasonal effects on stormwater concentrations. Table 8 shows the combined categorical groupings of the national data, with the limited seasonal distinctions. At the national level, EPA Rain Zones 1, 3, and 5 were found to have statistically significant differences in land use categories only for total suspended solids. EPA Rain Zones 1 and 2 were found to also have statistically significant differences in land use categories only for total suspended solids. Also, EPA Rain Zones 2 and 5 were not found to have statistically significant differences in land use categories, except for total copper. In addition, EPA Rain Zones 6 and 9 were not found to have statistically significant differences in land use groups, except for metals (total zinc and total copper). Table 8. Summary Table of Homogeneous Land Uses and Seasonal Clusters Stormwater Constituent Total Suspended Solids All EPA Rain Zones Land Use 1-RE,CO,ID 4-RE,CO,ID 6-RE,CO 9-CO,ID 2-RE 3-RE,CO,ID 2-CO,ID 5- RE,CO,ID 7- RE,CO,ID 9-RE 1-RE Total Zinc Total Copper 1-CO,ID 2-RE 3-RE,CO,ID 5-RE,CO,ID 2-ID 7-RE,CO,ID 9-RE,CO,ID 2-CO 4-RE,CO,ID 6-RE,CO 1-RE,ID 3-RE,CO,ID 5-RE,CO,ID 6-RE,CO 2-RE,ID 7-RE,CO,ID 1-CO 2-CO 9-RE,CO,ID 4-RE,CO,ID Mean (COV) 199 (1.9) 76 (1.6) 78 (1.9) 59 (1.9) 92 (1.6) 163 (2.3) 261 (1.2) 11 (2.3) 25 (1.9) 36 (1.2) 86 (1.9) 8 Total Phosphorous 1-CO 3-RE,ID 5-CO 1-RE,ID 2-RE,CO,ID 3-CO 4-RE,CO,ID 5-RE,ID 7-RE,CO,ID Total Kjeldahl Nitrogen Fecal Colifom 6-RE,CO 9-RE,CO,ID 1-RE,CO,ID 2-RE,CO,ID 3-CO,ID 4-RE,CO,ID 5-RE,CO,ID 7-(RE,CO) (FA,SU), ID 3-RE 7-(RE,CO) (SP,WI) 6-RE,CO 9-RE,CO,ID 1-(RE,CO,ID) (FA,SP,WI) 2-(RE,CO,ID) (SP,WI) 3-(RE,CO,ID) (SP,WI) 4-ID 7-RE,CO,ID 9-(RE,CO,ID) (SP) 1-(RE,CO,ID) (SU) 2-(RE,CO,ID) (FA,SU) 3-(RE,CO,ID) (FA,SU) 4-RE,CO 5-RE,CO,ID 6-RE 9-(RE,CO,ID) (FA,SU) 0.17 (1.2) 0.38 (1.7) 0.3 (1.2) 0.52 (0.67) 1.8 (0.99) 0.97 (0.90) 3.6 (0.73) 29120 (8.2) 40286 (3.0) 9